Hi, I’m Spencer. I build the whole data product, from the probabilistic model at its core to the app that ships it.
Background
I spent ten years at fast-moving companies like Uber and Gametime, building production ML and forecasting systems, running large-scale experiments, and turning messy data into decisions.
Since 2024 I’ve worked independently with startups and enterprise teams. Right now that means statistical modeling with VO Health and Bayesian demand forecasting with FieldGoal.
The work I like most sits where Bayesian modeling, agent tooling, and full-stack engineering meet: PyMC models that sample in the cloud, MCP servers that give AI agents real capabilities, and interactive apps built with React, Next.js, and Rust/WebAssembly.
Work
- Now
Statistical modeling · VO HealthBiomarkers for health and human performance, starting with a blood test for VO₂ max. - NowFieldGoalBayesian demand forecasting · FieldGoalHierarchical causal models of price elasticity and promotions for consumer packaged goods.
- 2021 – 2024
Senior Product Data Scientist · GametimeBuilt the experimentation stack behind company-wide A/B testing, and the scoring that surfaces the best ticket deals.
- 2019 – 2020
Senior Product Data Scientist · InVisionBuilt the next experimentation platform and used it to improve onboarding, retention, and conversion.
- 2017 – 2018
Data Science Curriculum Lead · DataCampDesigned and built the mobile data science curriculum, from the authoring format to the teaching guidelines.
- 2014 – 2016
Data Scientist · UberBayesian models of ride-sharing markets, Uber Eats dispatch optimization, and features for the internal ML platform.
Focus
01
Bayesian & probabilistic modeling
Designing and fitting models for inference, forecasting, and decisions under uncertainty.
PyMC · ArviZ · NumPyro · MCMC · hierarchical models
See cloudposterior, bayesDAG, Pacing, or the rest.
02
Production ML & forecasting
Forecasting and ML systems that run in production, and the experimentation frameworks that measure them.
scikit-learn · LightGBM · PyTorch · A/B testing · Snowflake · BigQuery
The day job at Uber, InVision, Gametime, and FieldGoal.
03
LLM & AI engineering
Agentic tooling and MCP servers, Claude API integrations, and outputs grounded with evals and citations.
Claude API · Model Context Protocol · Anthropic SDK · evals · RAG
See Pi Plugins, ArviZ MCP, Garmin MCP, or the rest.
04
Full-stack data products
Shipping end to end: data pipelines, APIs, and interactive front ends, then deploying and self-hosting them.
TypeScript · React · Next.js · Rust / WASM · PostGIS · Docker · Modal
See Effort Curve, GroundCover, Bikefitter, or the rest.
Education
- 2014
Master of Science in Artificial IntelligenceUniversity of San Francisco - 2011
Bachelor's in NeuroscienceRice University
Outside
Away from the keyboard I’m usually on a trail or a skin track: ski mountaineering, Nordic skiing, trail running, and the occasional triathlon. So far that adds up to 11,614 miles and 1,845,000 feet of climbing.


